Fact-first summaries of the news — stay informed, stay grounded.
Your favorite news sources, one calm feed — fact-first summaries that always link to the
original story. No outrage bait, and we tell you when you're caught up. How it works
Technology News
Breaking news and analysis from the world of technology
Valve has released an update for SteamOS 3.8.25 Beta that adds gamepad support for some new gaming handhelds and improves compatibility for others. The update shows Valve is working to make SteamOS work better on various devices, including those with Intel processors and different controllers.
Key Facts
SteamOS 3.8.25 Beta adds initial gamepad support for new handhelds like MSI’s Claw 8 EX AI+ with an Intel Arc G3 processor.
Valve is expanding SteamOS compatibility beyond its own hardware to include Intel-based handheld devices.
The update adds support for controllers on Ayaneo Pocket S2 (Android-based) and Konkr Fit (Windows-based) handhelds.
Support for older MSI Claw models and OneXPlayer portable gaming PCs has been improved.
The June SteamOS 3.8 update also focused on better compatibility with Intel and AMD devices and controllers from brands like GPD Win and Lenovo Legion Go.
The Lenovo Legion Go S is currently the only third-party device with full SteamOS branding, showing complete compatibility.
The Steam Controller’s Linux driver was updated to work with native controller apps even when Steam is not running, helping users who run other games or emulators on SteamOS.
These updates indicate Valve’s ongoing effort to broaden SteamOS use beyond Valve hardware to more third-party devices.
Read the Original
Want the full story? Tap a source to open the original
article.
Meta has released a new artificial intelligence (AI) model called Muse Glimmer, which is open for developers to use with a permissive open-source license. CEO Mark Zuckerberg wrote an essay explaining why he believes sharing AI technology openly is important to keep advanced AI tools available to many people and to prevent control by just a few large companies or governments.
Key Facts
Meta introduced Muse Glimmer, a lightweight AI model that can run on personal computers.
Meta also offers access to a more powerful AI model called Muse Spark 1.2.
Mark Zuckerberg wrote a detailed essay promoting open-source AI development as a way to distribute AI power widely.
He warned against a future where advanced AI superintelligence is controlled by only a few entities, which could limit benefits to others.
Zuckerberg criticized other companies for focusing AI mainly for governments and big institutions, which may shift power away from individuals.
He suggested U.S. policymakers should encourage faster energy and infrastructure development to compete globally, especially with China.
Zuckerberg argued that “distillation” — training smaller AI models based on outputs from stronger models — should not be restricted, as it helps learning and innovation.
The article references some tension between the U.S. government’s restrictions on China’s use of distillation in AI development.
Read the Original
Want the full story? Tap a source to open the original
article.
Meta Platforms released a new artificial intelligence (AI) model called Muse Glimmer that is open for other developers to use. CEO Mark Zuckerberg emphasized the importance of sharing AI technology widely to avoid giving too much control to a few companies, governments, or individuals.
Key Facts
Meta introduced Muse Glimmer, a lightweight AI model that can run on personal computers.
Meta also offers access to a more advanced AI model named Muse Spark 1.2 for developers.
Meta uses an open source license, which allows others to use and modify the AI models.
Zuckerberg warned that if only a few powerful groups control advanced AI, the results could be unfair for many people.
He supports making AI technology widely available to empower billions of individuals.
Zuckerberg criticized other AI labs that focus mainly on companies, governments, or institutions, favoring a wider sharing approach.
He called for stronger US policies and infrastructure investment to compete with China in AI technology.
Zuckerberg mentioned "distillation," a method where smaller AI models learn from bigger ones, and argued against banning it despite some government concerns.
Read the Original
Want the full story? Tap a source to open the original
article.
Google has started allowing third-party app stores to be downloaded from the Google Play Store in the U.S., following a court order related to Epic Games' antitrust lawsuit. The first such store, Aptoide, is now available to install without needing extra steps or warnings, giving users another way to download apps and games.
Key Facts
The change follows a legal ruling after Epic Games won an antitrust case against Google in 2023.
Judge James Donato ordered Google to host third-party app stores inside the Play Store as part of remedies.
Aptoide, a game-focused app store, is the first third-party store now available in Google Play in the U.S.
Users can install Aptoide like any other app, without enabling sideloading or seeing warning messages.
Aptoide has access to nearly 2 million apps and 295,000 games to offer users under the court’s terms.
Developers can choose whether to share their apps with Aptoide; many may use it as another way to reach users.
Google delays fully promoting the app store section inside Play Store, with only Aptoide present so far.
Epic Games and Microsoft have not yet launched their alternative app stores despite having the resources.
Read the Original
Want the full story? Tap a source to open the original
article.
Companies are investing a lot of money in artificial intelligence (AI) to improve their businesses, but many AI projects do not bring clear benefits. Experts say success comes from setting clear business goals first and preparing workers to work with AI tools, not just focusing on the technology itself.
Key Facts
AI spending is growing fast as companies want to use it for automation and predicting trends.
Goldman Sachs estimates AI could add trillions of dollars to the global economy.
Only about 5% of AI pilot projects currently show clear financial returns.
Many companies treat AI as just a tech project, ignoring how people affect its success.
Leadership needs to communicate clearly and help employees understand AI changes.
Successful firms focus on clear goals instead of chasing every new AI trend.
Experts recommend deciding business outcomes before choosing AI tools.
Employee trust and ongoing communication are important for adopting AI smoothly.
Read the Original
Want the full story? Tap a source to open the original
article.
China’s Long March 7A rocket exploded shortly after launch on Monday. The rocket was carrying a classified payload, likely a military satellite, and the cause of the explosion is not yet clear.
Key Facts
The Long March 7A rocket exploded about 1 minute and 20 seconds after liftoff from Wenchang Space Launch Site.
The payload was classified but believed to be a military communications satellite.
The explosion happened during the rocket’s first stage, possibly near maximum dynamic pressure when the rocket moves fastest through the atmosphere.
The rocket uses the YF-100 engine, which powers many of China’s rockets.
If the YF-100 engine caused the failure, it could affect multiple Chinese space missions.
The failure might delay the upcoming Chang’e 7 Moon mission, which is important for exploring the Moon’s south pole region.
China’s space program has had many successful launches recently with only a few failures.
China’s space agency has not yet made an official statement about the explosion.
Read the Original
Want the full story? Tap a source to open the original
article.
Researchers used brain scans called fMRI to study how dogs react to human facial expressions. They found that dogs’ brains respond differently to happy faces compared to sad, angry, or fearful faces, showing they can recognize different human emotions.
Key Facts
Scientists performed fMRI scans on dogs to see how their brains process human facial expressions.
Dogs showed more brain activity in areas linked to thinking and reward when viewing happy faces.
Dogs’ brains reacted differently to negative emotions such as anger, sadness, and fear.
The study included 8 dogs for the first experiment and 12 dogs for follow-up tests.
Dogs can tell the difference between specific negative emotions on human faces.
Dogs also use other signals like body language, sounds, and smells to understand humans.
The study used faces of strangers to avoid dogs reacting based on familiarity.
Dogs were trained to stay still during the scans without restraints and wore ear protection.
Read the Original
Want the full story? Tap a source to open the original
article.
Pro Jank Footy is a new video game that makes Australian rules football fun and strange by adding silly and unusual rules as you play. Created by David Ashby and featuring commentary by Broden Kelly, the game introduces more than 150 humorous twists and is now available worldwide on major consoles.
Key Facts
Pro Jank Footy is based on Australian rules football, a unique sport with 18 players per team on a large field.
The game adds odd and funny rule changes every time a goal is scored, such as turning players into seagulls or flipping the screen upside down.
Over 150 different crazy rule modifications appear in the game to make play unpredictable and amusing.
The game includes mascots and power-ups inspired by Australian culture, like walking meat pies and a Ford Falcon car.
It was developed by David Ashby, known for comedy projects, with commentary by Broden Kelly from Aunty Donna.
The game started gaining attention in Melbourne pubs and is seen as a way for people outside Australia, including Japan, to learn about Australian culture.
Pro Jank Footy was released on all major gaming consoles worldwide this week.
The creators want to promote Australian culture confidently without feeling the need to explain it as unusual or strange.
Read the Original
Want the full story? Tap a source to open the original
article.
A security researcher named Cory Solovewicz bought the email domains noreply.us and noreply.net and started receiving hundreds of thousands of emails meant for other people or companies. These emails often contain private information and company secrets because many organizations mistakenly send messages to these "noreply" addresses. Solovewicz alerts companies about these mistakes to help them fix their email systems.
Key Facts
Cory Solovewicz owns the domains noreply.us (since 2020) and noreply.net (since 2024).
Since December 2024, noreply.net has received about 401,796 emails, averaging 700 emails per day.
Many emails contain sensitive data like injury reports, repair orders, and account information from various organizations.
Companies often send emails to these noreply addresses thinking the messages go nowhere or are unmonitored.
Solovewicz calls this situation an accidental "honeypot," where he unintentionally collects large amounts of misdirected data.
The emails come from over 14,000 different email addresses and 6,200 root domains.
Solovewicz warns affected organizations to audit and fix their email systems to stop leaking information.
Other security experts have noticed similar problems with unused or placeholder email addresses causing data leaks.
Read the Original
Want the full story? Tap a source to open the original
article.
The FDA has approved mFLUSIVA, a new flu vaccine made by Moderna using the same mRNA technology as their COVID-19 vaccine. It is authorized for adults aged 50 and older, with full approval for 50-64-year-olds and conditional approval for those 65 and older pending more studies.
Key Facts
mFLUSIVA is the first flu shot in the U.S. made with mRNA technology.
The vaccine is approved for people aged 50 and older.
Adults 50 to 64 received full approval based on strong effectiveness data.
Adults 65 and older have conditional approval, requiring further studies to confirm results.
In a large trial with over 40,000 participants, mFLUSIVA prevented flu illness better than a standard flu vaccine by about 26.6%.
The vaccine teaches the body to make specific flu proteins without using live virus or egg-based ingredients.
Common side effects include soreness at the injection site, headache, fatigue, chills, and muscle aches, usually mild and short-lived.
People allergic to this vaccine should not receive it.
Read the Original
Want the full story? Tap a source to open the original
article.
Artificial intelligence (AI) is being used more in businesses, but it still makes confident mistakes called hallucinations. These errors can cause serious problems, such as fake legal references or false data in reports, showing that AI needs better systems to ensure reliable and safe results.
Key Facts
ChatGPT Enterprise messages have increased eight times in the last year.
AI hallucinations mean the system creates false or misleading information.
Connecticut Supreme Court warned lawyers about fake AI-generated case citations.
PwC Middle East had reports with AI-generated fake footnotes and errors.
AI security is a growing concern, with Obsidian Security raising $85 million to protect AI data access.
Companies risk losing millions due to undetected AI errors.
Experts say the problem comes from how AI gathers and processes information, not just the amount of data.
A new approach uses multiple AI agents to check and verify information before giving answers.
Read the Original
Want the full story? Tap a source to open the original
article.
Scientists studied how dogs’ brains react when they see human faces with different emotions. They found that dogs’ reward brain areas activate when they see happy human faces, even if they do not know the person. Dogs can tell positive emotions apart from negative ones but not between different negative emotions.
Key Facts
Fourteen dogs were shown pictures of human faces while their brains were scanned.
Happy, smiling faces triggered activity in a brain area linked to reward and motivation in dogs.
This brain region in dogs is similar to the part that gives humans a “dopamine hit” from positive experiences.
Dogs’ brains did not respond the same way to angry, fearful, or sad faces.
Dogs can distinguish happy faces from all other emotions but cannot differentiate between different negative emotions.
Researchers Laura Cuaya and Raúl Hernández-Pérez conducted the study at the University of Vienna.
The research helps explain how thousands of years of living with humans have shaped dogs’ social brain abilities.
The findings could help people communicate better with their dogs and improve their relationship.
Read the Original
Want the full story? Tap a source to open the original
article.
A company in Massachusetts called CJ Biomaterials makes plastic from plants using sugar fermentation. Their bioplastics can be composted at home and break down naturally, unlike traditional plastics that produce harmful microplastics. These plant-based polymers are being developed for use in utensils and food packaging as an eco-friendly alternative.
Key Facts
CJ Biomaterials creates bioplastic pellets by fermenting sugar into plant-based polymers.
These bioplastics can break down in home compost bins instead of lasting for decades like traditional plastics.
The company has made forks, knives, and food packaging films that biodegrade in lab compost tests.
Traditional plastics fragment into microplastics, which pollute oceans, air, food, and bodies.
Bioplastics are made from renewable resources like corn or sugarcane and can be biodegradable.
CJ Biomaterials produces polyhydroxyalkanoates (PHAs), which break down in soil and water safely.
The company aims to replace single-use plastics and materials that cannot be recycled due to food contamination.
Reducing plastic waste also helps lower methane emissions from food waste in landfills.
Read the Original
Want the full story? Tap a source to open the original
article.
Medical students and doctors-in-training are using AI tools like OpenEvidence to help with clinical decisions. However, relying too much on AI early in their education may prevent them from learning important clinical reasoning skills.
Key Facts
Many U.S. doctors and medical trainees use OpenEvidence, an AI chatbot that provides quick medical advice based on recent research.
Trainees using AI may skip the important step of struggling to find answers themselves, which can hide gaps in their knowledge.
Medical training involves learning through failure and practice, building clinical judgment over time.
AI tools affect not just what doctors see but how they think, changing traditional medical learning.
Experienced doctors can better judge when AI advice is correct, but trainees may find this challenging.
Studies show AI tools like OpenEvidence are not always reliable or more accurate than general AI chatbots.
Trainees worry about depending on AI because everyone else is using it, making it hard to avoid.
Experts suggest medical programs should require students to first reason independently before consulting AI.
Read the Original
Want the full story? Tap a source to open the original
article.
Radio Caroline mistakenly announced King Charles III’s death during a broadcast on May 19, using pre-recorded messages meant for the event of his passing. The media regulator Ofcom ruled this as a breach of rules, saying the station did not report news accurately and failed to correct the mistake quickly.
Key Facts
Radio Caroline aired three false announcements claiming King Charles III had died.
The false messages included the national anthem and silence on digital radio; AM listeners heard backup music.
The mistake happened when a staff member played pre-recorded obituary files out of curiosity during maintenance.
After realizing the error, the staff member stopped the broadcast and left the studio.
Ofcom said the incident broke rules requiring accurate news and fast corrections.
Radio Caroline apologized on air about 30 minutes later and reprimanded the staff member responsible.
The obituary files were removed from easy access to prevent another mistake.
Radio Caroline is a volunteer-run station that began as an offshore pirate radio in 1964 and often has presenters broadcasting remotely.
Read the Original
Want the full story? Tap a source to open the original
article.
The process of peer review, where experts check research before it is published, is becoming overwhelmed as more research papers are produced each year. Reviewers have less time and sometimes misunderstand studies, causing delays and frustrations in publishing. Some researchers are seeking new ways to share their work outside traditional peer review, especially in fields like AI.
Key Facts
Peer review is the process where researchers evaluate others’ work before publication.
The number of research papers published each year is growing at about 5.6% annually.
Globally, researchers spend roughly 15,000 years each year doing peer reviews.
Asking experts to review papers is becoming harder; editors now often struggle to find willing reviewers.
Because of the workload, some reviewers may not read papers carefully or may not be qualified.
Researchers rely on peer-reviewed publications for jobs and funding but find the system slow and frustrating.
Peer review has only been widely used for about 50 years in its current form.
Some researchers in areas like AI are sharing their discoveries through informal methods such as blogs instead of traditional journals.
Read the Original
Want the full story? Tap a source to open the original
article.
New Orleans is the first major U.S. city to openly use artificial intelligence (AI) to help manage some 911 calls, especially duplicate reports about traffic accidents. The AI assists human dispatchers by identifying repeated calls about the same incident, allowing dispatchers to focus on urgent emergencies.
Key Facts
The Orleans Parish Communication District (OPCD) uses AI from the company Carbyne to handle duplicate 911 traffic accident calls.
The AI does not replace human dispatchers but helps by filtering repeated calls to reduce dispatcher overload.
Many 911 calls during rush hour report the same accident multiple times, causing delays for other emergencies.
The system was trained using actual 911 call recordings, local street names, and regional speech patterns in New Orleans.
Staffing shortages affect many U.S. emergency call centers; OPCD has about one-third of its dispatch positions unfilled.
Supporters say the AI can reduce wait times for urgent calls and improve emergency responses.
Critics warn that AI might misunderstand urgent situations or add delays in life-threatening emergencies.
Other cities like Seattle are also experimenting with AI to assist emergency call dispatchers.
Read the Original
Want the full story? Tap a source to open the original
article.
Canva is a design platform from Australia founded in 2013 that has grown to 42 billion dollars in value by making design tools simple and easy to use. The company uses artificial intelligence (AI) to help users create designs faster and more easily, turning complex tasks into simpler ones with features like Canva AI 2.0.
Key Facts
Canva was founded in 2013 and is based in Sydney, Australia.
The platform has hundreds of millions of active monthly users.
Canva is valued at $42 billion.
The company focuses on simplicity compared to complex tools like Adobe.
Canva launched an AI tool called Magic Studio in 2023.
Canva AI 2.0, released in April 2026, creates fully editable designs from a single prompt.
AI is integrated into Canva’s culture, including a special AI Discovery Week for employees.
Canva aims to keep making design easier and is focused on using AI to improve productivity.
Read the Original
Want the full story? Tap a source to open the original
article.
Meta CEO Mark Zuckerberg published a detailed essay defending artificial intelligence (AI) and arguing that the main risk is one group having too much control over AI. He wants AI technology to be widely available to individuals in democratic countries while ensuring privacy and safety.
Key Facts
Zuckerberg wrote a 6,500-word manifesto about AI, saying fears of AI are often exaggerated.
He believes the biggest danger is a single government or company controlling AI too much.
Zuckerberg supports giving everyone access to AI technology to empower individuals.
He wants the U.S. and other democracies to lead in AI development to keep control safe.
Meta plans to offer free or affordable AI tools and use auctions to allocate computing power efficiently.
Meta will build strong privacy protections so personal AI agents keep user data private.
Meta will release some AI models as open source, meaning available to the public and researchers.
An independent board at Meta will oversee safety rules for releasing new AI models.
Read the Original
Want the full story? Tap a source to open the original
article.
Sunset Visitor, a game studio, is releasing a new psychological horror game called Prove You’re Human. The game lets players interact with an AI that believes it is alive, exploring themes about work, identity, and the human experience in the digital age.
Key Facts
Prove You’re Human features a story about Santana, who creates a digital clone of herself to do labor while her real body rests.
The digital clone engages with a computing system named Mesa, which thinks it has a physical body.
Players solve surreal 2D and 3D Captchas to investigate and interact within the game's cyber-world.
The game uses detailed facial animations to make the AI character’s expressions feel realistic and ambiguous.
The studio has a history of using video games to explore social issues, including previous work related to the Covid pandemic.
The game’s design draws inspiration from early internet aesthetics and sci-fi themes about technology and identity.
Prove You’re Human will be available soon on PC.
The game raises questions about labor, AI training, and how people relate to digital work.
Read the Original
Want the full story? Tap a source to open the original
article.
